The goal of this research is to develop and test a comprehensive model of the ways in which community context affects health across the life course. Specifically, communities characterized by social and physical disorder-by crime, unemployment, vandalism, danger of violent victimization, dirt, noise, and drugs--set in motion a destructive cycle of fear of victimization, mistrust, social isolation, a sense of powerlessness, distress and inactivity that ultimately affects health. Further, community context is hypothesized to have the greatest effect on the health of persons aged 60 and older. For example, fear of violent victimization is hypothesized to increase social isolation. First, people who are afraid mistrust others, which interferes with the establishment of social bonds. Second, fear directly increases social isolation as people who are afraid become more and more isolated in their house of apartment, rarely leaving to take a walk to the store or to visit neighbors. Older persons are likely to be especially susceptible to the processes by which community context affects health. A sedentary lifestyle, for example, in which the person rarely walks, coupled with social isolation, may be most detrimental to the health of older persons. Thus, community context is hypothesized to interact with age to affect health. Community context is the ultimate exogenous variable of interest. Health (measured as self-reports, physical dysfunction, conditions, and symptoms) is the final dependent variable. Variables mediating the effect of community on health (process variables) include victimization, fear of victimization, the sense of powerlessness, mistrust, social isolation, psychological distress (measured as symptoms of depression and anxiety), and physical activity. Age is the major conditioning (interacting, or moderating) variable of interest. A sample of 2,000 English speaking respondents in Illinois will be taken. Respondents will be interviewed by telephone at two points in time. An oversample of persons age 60 and over will be taken to ensure large enough numbers--approximately 880 persons over age 59, and 1,120 persons age 59 and younger. Modified random digit dialing will be used to ensure a representative sample. Psychometric analysis will be used to examine issues related to measurement of the concepts, and regression analysis, path analysis, structural equation models, and hierarchical linear models will be used to examine the proposed model.